GEOPHYSICAL RESEARCH LETTERS, VOL. 30, NO. 8, 1446, doi:10.1029/2003GL017126, 2003

Structure of oceanic core complexes: Constraints from seafloor gravity measurements made at the Atlantis Massif Scott L. Nooner, Glenn S. Sasagawa, Donna K. Blackman, and Mark A. Zumberge Scripps Institution of Oceanography, La Jolla, California, USA Received 3 February 2003; revised 25 March 2003; accepted 28 March 2003; published 30 April 2003.

[1] Using the DSV Alvin, the relative seafloor gravimeter are rafted to the surface. Therefore, fault rocks and ultramafic ROVDOG was deployed at 18 sites on the Atlantis Massif rocks that have been sampled on or near the Atlantis Massif (located at the ridge-transform intersection of the Mid- and other hypothesized core complexes [Dick et al., 1991; Atlantic Ridge and the Atlantis Transform Fault near 30N, Cannat et al., 1995; Karson, 1999; MacLeod et al., 2002; 42W). These data along with previously collected shipboard Tucholke and Lin, 1994; Cann et al., 1997; Tucholke et al., gravity and bathymetry provide constraints on the density 2001; Blackman et al., 2001] are congruent with this idea. structure of this oceanic core complex. A series of quasi 3-D [3] Seismic evidence of a detachment fault on the African forward models suggests that symmetric east and west- plate capable of exhuming lower crustal or upper mantle dipping density interfaces bound the core of the massif with material has been shown by Ranero and Reston [1999]. dip angles of 16–24 in the east and 16–28 in the west, Uplift of high-density rock in this way might explain the creating a wedge with a density of 3150–3250 kg/m3. The dip presence of the observed gravity high on the Atlantis angle in the east is steeper than that of the surface slope, Massif. However, Campbell and John [1996] gave evidence suggesting that the detachment fault surface does not coincide for emplacement of a synextensional dike-like pluton with the density boundary. The resulting low-density layer is beneath a detachment fault within the Colorado extensional interpreted as a zone of serpentinization. INDEX TERMS: corridor to explain the gravity high observed there. We use 3010 Marine Geology and Geophysics: Gravity; 3094 Marine new seafloor gravity data in combination with existing sea Geology and Geophysics: Instruments and techniques; 1219 surface gravity data to constrain a suite of 3-D forward Geodesy and Gravity: Local gravity anomalies and crustal models in order to place bounds on the geometry and value structure; 8150 Tectonophysics: Plate boundary—general (3040); of the anomalous density in the exposed oceanic core 9325 Information Related to Geographic Region: Atlantic Ocean. complex at Atlantis Massif. Citation: Nooner, S. L., G. S. Sasagawa, D. K. Blackman, and M. A. Zumberge, Structure of oceanic core complexes: Constraints 1.2. Seafloor and Sea Surface Gravity Acquisition from seafloor gravity measurements made at the Atlantis Massif, [4] We collected the seafloor gravity data with our sea- Geophys. Res. Lett., 30(8), 1446, doi:10.1029/2003GL017126, floor gravimeter, ROVDOG (Remotely Operated Vehicle 2003. deployable Deep Ocean Gravimeter), during the Nov.– Dec., 2000 cruise of the R/V Atlantis as one component of MARVEL 2000. The ROVDOG was carried outside DSV 1. Introduction Alvin and placed on the sea floor while an operator inside 1.1. Background the Alvin controlled the instrument and observed the data [2] This study was focused on Atlantis Massif, which is collection in real time. After an observation time of 10–20 located at the eastern inside corner of the intersection minutes the instrument was retrieved and transported to the between the Atlantis Transform Fault and the Mid-Atlantic next site. A seafloor survey taken with a single meter at an Ridge (MAR) at 30N, 42W. Spreading parallel corruga- average depth of 300 m has demonstrated a precision of 28 tions on the domal surface (Figure 1) are reminiscent of mGal or better. Further technical details may be found in surface morphology at some continental core complexes in Sasagawa et al. [2003]. the Basin and Range and have been interpreted as character- [5] The locations of the gravity dive sites (shown in istics of a detachment fault surface [Davis and Lister, 1988; Figure 1) form an approximately spreading-parallel line. Cann et al., 1997]. Similar features have been seen at two During each of the dives, the Alvin gathered rock samples fossil massifs along the Atlantis Fracture Zone [Blackman et and stopped for gravity measurements. On average, the 18 al., 1998; Cann et al., 1997], at other places along the MAR gravity sites were spaced 557 meters apart during each of [e.g., Cannat et al., 1995; MacLeod et al., 2002; Tucholke et the four dives1. The sea surface gravity data that we used as al., 2001] and elsewhere [e.g., Ranero and Reston, 1999]. part of the modeling are described in Blackman et al. The current accepted models are that these topographic highs [1998]. The uncertainty in these data are 1.8 mGal as are oceanic core complexes that form when the extension of indicated by the standard deviation in track crossing misfits. the crust is taken up by faulting along a detachment fault The location of the line is shown in Figure 1. rather than by plate accretion [Dick et al., 1991; Mutter and Karson, 1992; Tucholke and Lin, 1994]. As the crust extends via this low angle fault, lower crustal and upper mantle rocks 2. Data Reduction for Seafloor Gravity

[6] We corrected the gravity data for instrument drift, Copyright 2003 by the American Geophysical Union. tides, latitude, and lithospheric cooling. We also made free 0094-8276/03/2003GL017126$05.00 water corrections to the data. The resulting uncertainties are

29 -- 1 29 - 2 NOONER ET AL.: STRUCTURE OF OCEANIC CORE COMPLEXES

Table 1. Gravity Data Uncertainties Error Source Uncertainty (mGal) Gravimeter accuracy 0.028 Tides 0.010 Latitude correction 0.071 Free water correction 0.030 Bouguer slab correction 0.004 Terrain correction 0.262 Total RMS uncertainty 0.275

unmatched depth, a Monte Carlo approach was used: nor- mally distributed zero mean random noise (with standard deviation s = 20 m) was added to the bathymetric data and the terrain correction was computed. One thousand iterations of this yielded an average standard deviation of 0.262 mGal, which was adopted as the terrain correction uncertainty. This provides a better statistical estimate of uncertainty than simply computing the gravity effect of a 20 m slab, which would give an uncertainty value of about 2.4 mGal. [8] The complete Bouguer anomaly is shown in Figure 2. The RMS uncertainty in the measurements and corrections is summarized in Table 1. The total uncertainty (0.275 mGal) is Figure 1. Bathymetry map with the location of seafloor dominated by the imperfect terrain correction (0.262 mGal), gravity sites shown as white circles. Alvin dives are demonstrating the need for more detailed bathymetry in high numbered. Notice the corrugations running from east to precision seafloor gravimetry. However, the results of this west near dive 3642. study were not strongly influenced by this issue. summarized in Table 1. Further details can be found in 3. Gravity Modeling 1 Appendix 2 of the supporting material . The accuracy in [9] Although difficult to model, small-scale features calculating the complete Bouguer anomaly from seafloor undoubtedly affect our ability to interpret the data. For gravity data is greatly dependent on how well the shape of instance, there is probably a continuum of density values as the seafloor itself is known. In a region like the MAR 30 N well as a complicated density interface boundary, such as a area, the terrain is often rugged and steep, making a precise steepening or shallowing dip angle with depth. Due to these terrain correction difficult. The regional bathymetry was factors and to the inherent non-uniqueness of gravity, we gridded with a 100 m spacing [Blackman et al., 1998], approached the problem by looking only at simple, end which effectively smoothes the terrain, causing discrepan- member model geometries that neglect these second order cies between the actual depth and that predicted by the grid. effects. Thus we tested four types of models: homogeneous To minimize this problem, depth was measured at the density, one-fault, cylindrical plug, and wedge. gravity sites with a Paroscientific 410k pressure gauge which has demonstrated a repeatability of 3 cm in relative depth [Sasagawa et al., 2003]. The free water corrections and the slab component of the terrain correction were made using these measured instrument depths. The resulting uncertainty in the slab correction is 0.004 mGal, using 2900 kg/m3 as the reference density. We estimate the uncertainty in the free-water anomaly to be 0.083 mGal, whereas previous seafloor gravity studies reported uncer- tainties of 0.30 mGal or greater [Luyendyk, 1984; Hilde- brand et al., 1990; Holmes and Johnson, 1993; Ballu et al., 1998; Cochran et al., 1999]. [7] The RMS difference between the measured depths and those obtained from the bathymetry grid was 37 meters (150 m maximum). Shifting the survey points 100 meters west relative to the bathymetry reduced the maximum difference to about 30 m, with an RMS difference of 20 m. To determine the error in the terrain correction due to this

1Supporting material is available via Web browser or via Anonymous Figure 2. This figure shows one model fit to the seafloor FTP from ftp://ftp.agu.org, directory ’’‘‘ apend(Username = ‘‘anony- mous’’, Password = ‘‘guest’’); subdirectories in the ftp site are arranged by and sea surface data. The dip angles bounding the high- paper number. Information on searching and submitting electronic supple- density wedge are 16 in the east and 20 in the west; Ár is 3 ments is found at http://www.agu.org/pubs/esupp_about.html. 250 kg/m . NOONER ET AL.: STRUCTURE OF OCEANIC CORE COMPLEXES 29 - 3

[10] Blackman et al. [1998] observed that the gravity high is slightly east of the topographic high at the Atlantis Massif. Our seafloor gravity measurements confirm this. Therefore a homogeneous density model does not adequately fit both data sets, since the gravity follows topography in such a case. However, the best fitting density to this model was useful in justifying our choice for the reference density used in the Bouguer and terrain corrections (2900 kg/m3). A one- fault model with one east dipping density boundary has the same difficulty as the one density model, indicating that a west-dipping boundary may be present. We also modeled a vertically oriented cylindrical plug of higher density cen- tered on the gravity high, as has been documented in the Basin and Range [Campbell and John, 1996]. However, for the Atlantis Massif, this geometry requires an extremely large density contrast (greater than 1200 kg/m3) and a radial extent of 10–20 km for even a nominal fit to the seafloor data1. The best fitting simple model that we examined consisted of only two bodies: a high-density wedge and the surrounding terrain. The boundaries of the high-density Figure 4. Standard deviation,s, of the residuals to the wedge are sides sloping off to the east and to the west model fit as a function of the eastern boundary dip angle (Figure 3) and the transform fault to the south. The northern and density contrast with respect to the 2900 kg/m3 boundary is at 30200N. Tests showed that the geometry of reference. For this figure, the dip is constant at 20 in the the north and south boundaries had no significance on the west. Seafloor results are shown by solid lines and sea model results. We swept through all combinations of east surface results are shown by dashed lines. The models that and west dip angles and wedge density contrast, requiring fit both data sets best, shaded region, have central high that the model fit both sea-surface and seafloor data. density wedge boundaries with dip angles between 16–24 [11] Figure 4 shows the standard deviation, s, of both the in the east and 16–28 in the west and density contrasts of seafloor and the sea surface residuals to the wedge model as 250–350 kg/m3. a function of the east boundary dip angle and density contrast with respect to a 2900 kg/m3 reference (the dip angle in the west is held fixed at 20 for the figure). The at which the corrugated seafloor dips (11), indicating that geometries that fit both data sets best have boundary dip the density boundary does not coincide with the fault angles between 16–24 in the east and 16–28 in the west surface. It is worth noting that the seafloor data appear to and density contrasts of 250–350 kg/m3. These ranges were constrain the model more tightly than the sea surface data, obtained by requiring s < 2 mGal for both seafloor and sea despite the limited coverage. The seafloor data favor higher surface. For the sea surface, this range was chosen because density for the wedge and a low angle boundary dip, while it is about the same as the uncertainty in the data. It was the sea surface data favor geometries with lower density chosen for the seafloor because 2 mGal is approximately the contrast and do not constrain the fault angles well. How- small-scale variation of the data. This was determined from ever, a combination of the two data sets provides much the residuals to a best-fitting second order polynomial. The better constraints than either alone. range of eastern dip angles is slightly steeper than the angle 4. Discussion

[12] At the Atlantis Massif, most of the detachment fault surface is draped with pelagic ooze of some thickness, making outcrops of basement rock difficult to find. How- ever, serpentinized peridotites and evidence of shearing are widespread across the exposed southern wall of the massif [Blackman et al., 2001]. Seismic data also suggest that partially serpentinized to unaltered peridotite exist less than 1 km below the seafloor in this area [Collins et al., 2001]. Consistent with this, our gravity results indicate that the core of the massif is composed of rocks with an average density of 3150–3250 kg/m3. The best fitting geometries Figure 3. The boundaries of the high-density area form a indicate that both the east and west boundaries are more wedge with sides sloping off to the east and to the west. We steeply dipping than the corrugated surface of the massif, as computed a suite of 3-D forward models, varying the dip hinted at by the eastward shift of the gravity high with angle and the density contrast of the wedge, requiring that respect to the topographic high. This means that in the east the model fit both sea-surface and seafloor data. This wedge the inferred detachment fault surface does not coincide with terminates at a depth of 6000 m below sea level (base the density boundary, creating a strip of less dense material depth). The horizontal extent of this region is 47.459 km up to 1 km thick within the foot wall block. This is also seen and the vertical extent is 5 km. in the results of the NOBEL seismic experiment [Collins et 29 - 4 NOONER ET AL.: STRUCTURE OF OCEANIC CORE COMPLEXES al., 2001]. This low-density layer is most likely an alteration Blackman, D. K., J. R. Cann, B. Janssen, and D. K. Smith, Origin of front that is sub-parallel to the detachment fault surface. extensional core complexes: Evidence from the Mid-Atlantic Ridge at Atlantis Fracture Zone, J. Geophys. Res., 103, 21,315–21,333, 1998. Supporting this idea, rock samples from the detachment Blackman, D. K., D. S. Kelley, J. A. Karson, and MARVEL2000, New surface indicate that it is composed primarily of serpenti- seafloor maps and samples from the Mid-Atlantic Ridge 30N oceanic nized peridotite with some lesser amount of gabbroic core complex, Eos Trans. AGU, 82(47), Fall Meet. Suppl., 2001. Campbell, E. A., and B. E. John, Constraints on extension-related pluton- material [Cann et al., 1997]. A less likely scenario is that ism from modeling of the Colorado River gravity high, Geolog. Soc. Am. this zone is instead magmatic in nature; i.e. rotated volcanic Bulletin, 108, 1242–1255, 1996. intrusives created during a brief magmatic phase early in the Cann, J. R., D. K. Blackman, D. K. Smith, E. McAllister, B. Janssen, S. Mello, E. Avgerinos, A. R. Pascoe, and J. Escartin, Corrugated slip evolution of the massif, and later cut by the detachment surfaces formed at ridge-transform intersections on the Mid-Atlantic fault, uplifted, and exposed at the surface. However, our Ridge, Nature, 385, 329–332, 1997. modeling shows that large-scale dike-like intrusions beneath Cannat, M., C. Mevel, M. Maia, C. Deplus, C. Durand, P. Gente, P. Agri- the detachment surface are not the cause of the observed nier, A. Belarouchi, G. Dubuisson, E. Humler, and J. Reynolds, Thin crust, ultramafic exposures, and rugged faulting patterns at the Mid- gravity anomaly. Instead, the observed wedge-like geometry Atlantic Ridge (22–24N), Geology, 23, 49–52, 1995. of the core is consistent with the unroofing of deep seated Cochran, J. R., D. J. Fornari, B. J. Coakley, R. Herr, and M. A. Tivey, rock by extension and rotation along a detachment fault. Continuous near-bottom gravity measurements made with a BGM-3 gravimeter in DSV Alvin on the East Pacific Rise Crest near 9310N [13] The western density boundary also dips below the and 9500N, J. Geophys. Res., 104, 10,841–10,861, 1999. seafloor. Several ideas can be put forth to explain this. First, Collins, J., and R. S. Detrick, Seismic structure of the Atlantis Fracture this could be a compositional boundary due to layering of a Zone megamullion, a serpentinized untramafic massif, Eos Trans. AGU, 79, F800, 1998. rotated crustal block. Second, the low-density region might Collins, J. A., B. E. Tucholke, and J. Canales, Structure of Mid-Atlantic be due to the presence of a rider block that was carried Ridge megamullions from seismic refraction experiments and multichan- eastward during extention along the detachment. Finally, the nel seismic reflection profiling, Eos Trans. AGU, 82, Fall Meet. Suppl., zone could be a continuation of the alteration front asso- F1100, 2001. Davis, G. A., and G. S. Lister, Detachment faulting in continental extention; ciated with the detachment fault surface. The breakaway perspectives from the southwestern U.S. cordillera, Spec. Pap. Geol. Soc. zone is thought to be several kilometers west of the modeled Am., 218, 133–159, 1988. area. Our limited gravity coverage in the west does not Dick, H. J. B., P. S. Meyer, S. Bloomer, S. Kirby, D. Stakes, and C. Mawer, Lithostratigraphic evolution of an in-situ section of oceanic layer 3, in allow us to make any conclusive statements regarding Von Herzen and Robinson et al., Proc. ODP Sci. Res., 118, 439–538, structure of this area. 1991. [14] These interpretations are consistent with the presence Hildebrand, J. A., J. M. Stevenson, P. T. C. Hammer, M. A. Zumberge, and R. L. Parker, A seafloor and sea surface gravity survey of Axial Volcano, of a detachment fault. Although some seismic studies have J. Geophys. Res., 95, 12,751–12,763, 1990. been done in the region [Collins and Detrick, 1998; Collins Holmes, M. L., and H. P. Johnson, Upper crustal densities derived from et al., 2001], more detailed results could help further seafloor gravity measurements: northern Juan de Fuca Ridge, Geophys. constrain dip angles and elastic properties of the massif, Res. Lett., 20, 1871–1874, 1993. Karson, J. A., Geological investigation of a lineated massif at the Kane allowing us to better define the density contrast and geom- Transform Fault: implications for oceanic core complexes, Phil. Trans. etry. Because of the limited coverage of our single-profile Roy. Soc. Lond., A, 357, 713–736, 1999. seafloor data set it is not beneficial to test more complicated Luyendyk, B. P., On-bottom gravity profile across the East Pacific rise crest at 21 north, Geophysics, 49, 2166–2177, 1984. models, although it is likely that there is some small scale MacLeod, C. J., J. Escartin, D. Banerji, G. J. Banks, M. Gleeson, D. H. B. surface density variability as well as subsurface density Irving, R. M. Lilly, A. M. McCaig, Y. Niu, S. Allerton, and D. K. Smith, structure. In fact, because of the limited extent, the gravity Direct geological evidence for oceanic detachment faulting: The Mid- Atlantic Ridge, 15450N, Geology, 30, 879–882, 2002. data provide little north-south constraints on the structure. Mutter, J. C., and J. A. Karson, Structural processes at slow spreading These come mostly from the morphology of the region. ridges, Science, 257, 627–634, 1992. Further studies of these structures with more extensive Ranero, C. R., and T. J. Reston, Detachment faulting at ocean core com- gravity measurements, seismic studies, and drilling will plexes, Geology, 27, 983–986, 1999. Sasagawa, G., W. Crawford, O. Eiken, S. Nooner, T. Stenvold, and help us to understand these interesting extensional regions. M. Zumberge, A new seafloor gravimeter, Geophysics, in press, 2003. Tucholke, B. E., and J. Lin, A geological model for the structure of ridge [15] Acknowledgments. We thank the shipboard participants on segments in slow spreading ocean crust, J. Geophys. Res., 99, 11,937– cruise AT03-60, the officers, crew and pilots of the RV Atlantis and DSV 11,958, 1994. Alvin, Statoil for the loan of ROVDOG, Robert L. Parker for software, and Tucholke, B. E., K. Fujioka, T. Ishihara, G. Hirth, and M. Kinoshita, Sub- Eric Husmann for instrument preparation, and the reviewers for their mersible study of an oceanic megamullion in the central North Atlantic, suggestions. J. Geophys. Res., 106, 16,145–16,161, 2001. References Ballu, V., J. Dubois, C. Deplus, M. Diament, and S. Bonvalot, Crustal structure of the Mid-Atlantic Ridge south of the Kane Fracture Zone D. K. Blackman, S. L. Nooner, G. S. Sasagawa, and M. A. Zumberge, from seafloor and sea surface gravity data, J. Geophys. Res., 103, Scripps Institution of Oceanography, La Jolla, CA 92093-0225, USA. 2615–2631, 1998. ([email protected])